The molecular pathology of gastric carcinoma, the second most common cancer in the world, and of its precancerous precursor lesions has become much better elucidated in recent years, but much remains to be discovered. For example, tumor suppressor gene mutation, DNA hypermethylation, microsatellite instability (MSI), and altered gene expression have been shown to occur frequently during the origin and progression of this cancer. However, a greater understanding of these mechanisms and their timing in gastric carcinogenesis could lead to the ultimate development of early detection and staging biomarkers, as well as novel therapeutic approaches to this deadly disease. Therefore, in order to deepen our understanding of molecular mechanisms underlying gastric carcinogenesis, we will pursue the following Specific Aims: 1) To compare and contrast the global gene expression patterns of various subtypes and premalignant stages of gastric cancer, using cDNA microarrays and bioinformatics algorithms. 1a) cDNA microarrays will be produced and hybridized to amplified RNA from various histologic subtypes and premalignant stage precursor lesions of gastric cancer. 1b) The bioinformatics programs Cluster/Treeview, significance analysis of microarrays (SAM), artificial neural networks (ANNs), and support vector machines (SVMs) will be used to derive global gene expression patterns for each subtype of gastric cancer or precursor lesion. During this process, we will also identify genes that are significantly differentially expressed among the various subgroups. 2) To perform a limited methylotyping assay to ascertain epigenetic signatures of different types of gastric cancer and premalignant lesions, focusing on known as well as novel CpG islands in the promoter or upstream regions of genes. 2a) Known methylation targets will be analyzed, including E-cadherin (CDH 1), p 16, p 15, p 14-ARF, death-associated protein kinase (DAPK), O6-methylguanine DNA methyltransferase (MGMT), human mutL homolog 1 (hMLH1), adenomatous polyposis coli (APC), RASSFIA, deleted in colon carcinoma (DCC), and 14-3-3-sigma. 2b) Novel methylation target searches will be performed using CpG island microarrays, with alternative backup approaches to include methylation-sensitive genomic DNA digestion followed by arbitrarily primed PCR (AP-PCR), as well as single-gene MSP. 3) To correlate the results of, aims (1) and (2) with each other as well as with other clinical and biologic parameters of these gastric lesions.